A forecast you never grade is just an opinion you got to feel smart about. So here's the full accounting after three months — the hits, the misses, and the one structural thing I got wrong in a way that's worth more than any of the things I got right.
Where we are
The shape of the war has changed. Project Freedom's offensive phase wound down (Operation Epic Fury concluded in early May). A negotiated framework is now on the table — an MOU under which Iran keeps nominal sovereignty over the strait but gives up tolls and clears its mines in exchange for sanctions relief. Netanyahu publicly thanked Trump for it, so Israel's on board. Pezeshkian opened the door on the nuclear file. Trump's posture shifted to "time is on our side, not rushing."
And the politics got loud. GOP hawks — Cruz, Wicker, Graham, Pompeo, Bolton, Massie — are panning the emerging deal as a "disastrous betrayal," angry that Iran keeps sanctions relief, enrichment capability, and effective control of Hormuz. Republicans called off a vote on the war resolution on May 21 because it was about to pass. Tulsi Gabbard resigned as DNI on May 22.
Markets read the de-escalation: Brent fell from ~$116 on May 1 to ~$104 by May 22; WTI dropped below $91. Gas is still painful — national average $4.55, a four-year high and up ~42% year over year, with California at $6.13, six states above $5, and nineteen states expected to post record Memorial Day prices. The price relief lags the diplomatic relief, which is normal.
What I got right
- "Trump is the more desperate party, not Iran." This held. He absorbed the May UAE strikes and pushed for a deal rather than retaliating into a wider war. The asymmetry call was the load-bearing one and it carried.
- The Scenario B / Project Freedom escalation. I weighted re-escalation at the top of the distribution; it hit on schedule. The branch I said was most likely is the branch that happened.
- The independent-reporting lead time. Through the conflict phase, independent outlets read every major turn 48–72 hours ahead of the corporate wires. Confirmed repeatedly.
What I got wrong
- "Iran won't yield on the nuclear program." Wrong. Pezeshkian publicly opened that exact door. I'd treated Iran's nuclear posture as a fixed constraint; it was a negotiating position.
- "This doesn't resolve until the US–Israel relationship breaks, or Iran fragments, or the US accepts Iranian control of the strait." Too strong. I'd framed the exit conditions as three hard, binary breaks. The actual path is a softer compromise — the MOU threads between them. Reality found a middle option I'd written off.
- The negotiated-de-escalation window. I underweighted it in the 2–4 week band. I had the escalation timing right and the de-escalation timing wrong, which is a specific, repeatable error: I was good at calling the breakdowns and bad at calling the settlements.
- One thing the independent sourcing got wrong too: its lead-time advantage inverted at the negotiation phase. The deal terms broke through corporate access journalism first — Axios had the MOU framework before the independent outlets. The edge was real for the war and gone for the diplomacy. Worth remembering that a source's advantage is domain-specific, not permanent.
The clock I misread
Here's the miss that actually taught me something. I modeled Iran's staying power on its military clock — the "six months of munitions" claim, which was real. What I underweighted was Iran's economic clock. Quincy Institute analysis put something like half of Iranian jobs at risk under the blockade economics. The economic pressure compressed Iran's political timeline faster than the military pressure did.
So the deeper read I missed: Iran's economic clock was shorter than its military clock. It could fight for six months and still need a deal in eight weeks. I was watching the wrong stopwatch. When I model a state's resolve next time, the question isn't just "can they keep fighting" — it's "can the people running it survive the domestic economics of fighting that long." Those are different numbers, and the smaller one wins.
The updated 30-day distribution
| Scenario | Weight |
|---|---|
| MOU signed in 1–2 weeks; oil into the $80s; gas back to $3.75–4 by July | ~50% |
| MOU language slips; Iran balks on nuclear; talks extend without a deal | ~25% |
| Deal collapses; hawk pressure forces re-escalation | ~15% |
| Iran signs but opens the strait selectively / slow-walks mine clearing | ~10% |
The center of gravity has moved from "how hot does this get" to "does the paper actually get signed and honored." That's a different question, and it's a better problem to have than the one we had in March.
What I'm watching
- Signed MOU text — not framework language, actual text. Until then the ~25% slip case is live.
- Mine clearing as the verification proxy. Iran can sign and still not open the strait; watch the minesweepers, not the press releases.
- Whether the hawk revolt has the votes to derail ratification on the U.S. side.
The meta-lesson
Three months, one running model, graded out loud. The headline asymmetry call held. The structural calls were too rigid — I framed exits as hard breaks when reality had a soft middle. And the single most useful correction was learning to separate a regime's military endurance from its economic endurance and to price the shorter one.
That's the whole point of writing these down. The wins feel good; the losses are where the model actually improves.
Final entry in this phase of the series. I'll keep updating as the MOU resolves — or doesn't.